CIVIL LAW AND UAE: PREDICTIVE HARM PREVENTION REPLACING EX-POST LIABILITY

1. Introduction

Traditional civil liability is predominantly ex-post.

This means:

Harm occurs first → liability is established later → compensation or another remedy follows.

For example:

Defective product → injury → lawsuit → proof of fault/causation → damages.

Predictive harm prevention proposes a different approach:

Potential harm detected → risk predicted → preventive intervention → harm avoided or reduced.

This approach becomes particularly important in:

artificial intelligence;

autonomous systems;

digital platforms;

financial technology;

cybersecurity;

medical technology;

smart buildings;

connected vehicles;

predictive analytics;

algorithmic decision-making;

automated contracting.

The central question is therefore:

Can UAE civil law move from merely compensating harm after it occurs toward preventing foreseeable harm before it occurs?

The answer under present UAE law is not a complete replacement of ex-post liability. Rather, preventive mechanisms can complement traditional liability. The current Civil Transactions Law continues to recognise compensation for harm, while courts can in appropriate circumstances provide protective or restorative remedies. The new Civil Transactions Law, Federal Decree by Law No. 25 of 2025, entered into force on 1 June 2026 and repealed Federal Law No. 5 of 1985.

Thus, the better legal model is:

Ex-post liability + Ex-ante prevention = Modern UAE civil protection

2. Meaning of Predictive Harm Prevention

Predictive harm prevention means using:

AI;

machine learning;

statistical models;

historical legal data;

sensor data;

risk scoring;

automated warnings;

predictive maintenance;

fraud detection;

anomaly detection;

digital monitoring;

to identify a substantial risk of future harm and intervene before the harm occurs.

Example

Suppose an autonomous industrial machine detects:

abnormal vibration;

overheating;

unusual movement;

repeated software errors.

A predictive system may determine that the machine has a high probability of causing an accident.

Instead of waiting for:

Accident → Injury → Compensation claim

the system could produce:

Risk detected → Machine stopped → Inspection → Accident prevented

This represents the basic philosophy of predictive harm prevention.

3. Ex-Post Liability

Traditional civil liability operates after a legally relevant event.

The usual sequence is:

Wrongful act → Damage → Causation → Liability → Compensation

The claimant normally needs to establish the legally relevant elements of the claim.

Under the current UAE Civil Transactions Law, Article 245 establishes the general civil-liability principle that a person causing harm may be required to compensate for that harm.

The new Code also contains provisions concerning the assessment of compensation and remedial measures. Article 255 addresses assessment according to the extent of loss, while Article 256 allows, depending on the circumstances and upon the injured party's request, restoration of the previous position or performance of a specific matter related to the harmful act in lieu of purely monetary compensation.

Therefore, UAE civil law remains fundamentally capable of responding to completed harm.

4. Why Ex-Post Liability May Be Insufficient

Compensation cannot always make the injured party whole.

Consider:

A. Death

Money cannot restore life.

B. Permanent disability

Compensation may alleviate consequences but cannot restore the original physical condition.

C. Environmental damage

Some environmental harm may be irreversible.

D. Data destruction

Once confidential data is disclosed, subsequent damages may not fully repair the loss.

E. Cybersecurity attacks

A financial loss may be compensated, but stolen information may remain exposed.

F. Digital assets

Once assets are transferred through multiple wallets, recovery can become extremely difficult.

G. AI decisions

A discriminatory or harmful automated decision may affect thousands of people before the problem is discovered.

This creates a fundamental principle:

Where harm is potentially irreversible, prevention can be more valuable than compensation.

5. The Preventive Model

A predictive civil-justice model may operate as:

Risk identification
↓
Probability assessment
↓
Early warning
↓
Preventive intervention
↓
Monitoring
↓
Human review
↓
Harm avoided/reduced

The legal objective changes from:

“Who must pay after the harm?”

to:

“Who had the ability and responsibility to prevent the foreseeable harm?”

6. Foreseeability as the Bridge

Foreseeability provides the conceptual connection between traditional liability and predictive prevention.

If a person can reasonably foresee that conduct may cause harm, the legal system may have stronger reasons to require reasonable precautions.

A DIFC Court of First Instance decision, Faizal Babu Moorkath v Expresso Telecom Group Ltd [2023] DIFC CFI 008, described duty of care in terms of reasonable foreseeability, proximity and whether it is fair, just and reasonable to impose the duty. The Court also explained that a duty of care involves taking reasonable care to avoid foreseeable harm.

This is highly relevant to predictive harm prevention.

Formula

Foreseeable Risk → Reasonable Precaution → Reduced Harm

7. Predictive Prevention Does Not Mean Automatic Liability

An important distinction must be made.

Suppose an AI system predicts:

“There is a 70% chance of future harm.”

That prediction alone should not automatically create liability.

The legal system must still ask:

Was the risk reasonably foreseeable?

Who knew or should have known about it?

Who had control over the risk?

Was preventive action reasonably available?

Was the cost of prevention proportionate?

Did the person have a legal duty to act?

Did failure to act cause the eventual harm?

Therefore:

Prediction ≠ Duty ≠ Breach ≠ Liability

These are separate legal questions.

8. Predictive Prevention and the Current UAE Civil Transactions Law

The current Civil Transactions Law entered into force on 1 June 2026 and replaced the 1985 Code.

This is important for legal analysis because many older UAE judgments refer to the former Civil Code's article numbers.

The new Code's civil-liability provisions now appear in a different numbering structure. Contemporary legal commentary identifies Articles 245–258 as the new framework for civil liability and compensation.

Therefore:

Historical cases remain useful for legal reasoning, but their old statutory article numbers must not automatically be treated as current article numbers.

9. Predictive Harm Prevention and Injunctions

One of the clearest existing legal mechanisms resembling predictive prevention is interim injunctive relief.

The court does not necessarily wait until the final damage occurs.

Instead, it may restrain conduct where the circumstances justify immediate protection.

This is structurally similar to predictive harm prevention:

Potential future harm → judicial intervention → prevention

10. Case Law 1: Golden Sands Hotel LLC v Brighton Rock Restaurant LLC — DIFC CFI 106/2025

In Golden Sands Hotel LLC trading as Hilton Dubai The Walk v Brighton Rock Restaurant LLC [2026] DIFC CFI 106, the DIFC Court considered an application for interim injunctive relief preventing the defendant from accessing or operating in part of the hotel pending trial.

The Court granted an interim injunction.

It considered:

whether there was a serious issue to be tried;

balance of convenience;

adequacy of damages;

practical consequences;

preservation of the claimant's position pending trial.

The Court found damages inadequate in the circumstances and granted interim protection.

Importance

This case demonstrates that civil justice does not always have to wait for final damages.

The court may intervene to prevent continuing or future harm.

Principle

Where monetary compensation may be inadequate, preventive judicial relief may be appropriate before final adjudication.

This is conceptually close to predictive harm prevention.

11. Case Law 2: Faizal Babu Moorkath v Expresso Telecom Group Ltd — DIFC CFI 008/2023

The Court discussed the concept of duty of care.

It explained that foreseeability of loss is an important factor and that a duty involves taking reasonable care to avoid foreseeable harm.

The case concerned alleged failures relating to financial transactions and verification.

Importance for predictive prevention

The case demonstrates the importance of asking:

Could the relevant risk reasonably have been anticipated and addressed?

That is precisely the question that predictive technology seeks to improve.

Predictive model

Transaction anomaly → warning → verification → intervention → loss avoided

instead of:

Transaction → loss → lawsuit → damages

12. Case Law 3: Haya Spa LLC v Harper Real Estate / Hasan Real Estate — DIFC SCT 150/2016

In Haya Spa LLC v Harper Real Estate / Hasan Real Estate [2016] DIFC SCT 150, the DIFC Court discussed negligence through:

duty;

breach;

causation;

damage.

The Court also considered foreseeability and the contribution of the claimant's own negligence.

The case illustrates the importance of foreseeable harm in determining whether a duty of care exists.

Importance

Predictive harm prevention can be understood as operationalising the foreseeability element.

Instead of merely asking after the accident:

“Was the harm foreseeable?”

a predictive system attempts to identify that foreseeable risk beforehand.

Principle

Foreseeability should encourage reasonable precautions, not merely retrospective compensation.

13. Case Law 4: Gate Mena DMCC v Tabarak Investment Capital Ltd — DIFC CA 002/2023

The Gate Mena litigation involved cryptocurrency and digital-asset issues.

The DIFC Court of Appeal considered duty-of-care questions and the relationship between foreseeability, proximity and responsibility. The case illustrates how conventional civil-liability concepts must be applied to technologically complex environments.

Importance

Digital assets create risks that can move extremely quickly.

For example:

Wallet compromise → rapid transfer → multiple transactions → difficult recovery

A purely ex-post compensation system may encounter substantial practical difficulties.

Predictive monitoring could instead identify:

unusual transaction patterns;

suspicious transfers;

abnormal wallet activity;

sudden changes in transaction behaviour.

Principle

Technology can move the legal system from post-event recovery toward early risk detection, but legal responsibility remains fact-specific.

14. Case Law 5: AES Middle East Insurance Broker LLC v GSB Capital Ltd — DIFC CFI 060/2023

The AES litigation involved massive electronic disclosure.

More than two million documents were processed, and an AI-driven application was used to identify potentially relevant images before potentially relevant documents were manually reviewed.

Importance

This case is not itself a predictive-liability judgment.

However, it demonstrates an important technological model:

AI identification → human verification → legal action

rather than:

AI prediction → automatic legal conclusion

Application to harm prevention

The same model can be used to identify risk:

AI detects anomaly → human reviews → preventive action

This is a practical example of human-controlled predictive systems.

15. Case Law 6: Klesta Eshja & Hair Creators Salon LLC v Salah Masri & Others — DIFC CFI 066/2024

In Klesta Eshja & Hair Creators Salon LLC v Salah Masri & Others [2024] DIFC CFI 066, AI was used substantially in preparing legal material, but the resulting material contained false references and misleading material.

The DIFC Court imposed procedural consequences and demonstrated that parties remain responsible for material placed before the court.

Importance for predictive prevention

The case demonstrates a crucial limitation:

Predictive systems themselves can become a source of legal harm.

Therefore, preventive AI must itself be subject to:

verification;

audit;

human supervision;

quality control;

accountability.

The lesson is not simply “use more AI.”

It is:

Use AI with controls capable of preventing AI-generated harm.

16. Case Law 7: Stelian Gheorghe v BSA Ahmad Bin Hezeem & Associates LLP — DIFC CFI 045/2025

In Stelian Gheorghe v BSA Ahmad Bin Hezeem & Associates LLP [2026] DIFC CFI 045, the Court was confronted with material that was alleged to have been partly generated by AI and contained legal errors.

The Court observed that errors of law have no place in witness evidence filed by lawyers and ultimately stayed the proceedings in favour of arbitration pursuant to the arbitration agreement.

Importance

This case illustrates:

AI-generated legal content can contain errors;

technological assistance does not remove professional responsibility;

legal systems need safeguards against AI-generated procedural harm.

Principle

Preventive AI governance must protect the legal system from the risks created by AI itself.

17. Case Law 8: Niran v Nysa — DIFC SCT 403/2023

In Niran v Nysa [2023] DIFC SCT 403, the Court considered employer responsibility and preventive obligations in the context of workplace conduct.

The applicable DIFC employment provision contemplated employer liability where the employer failed to take reasonably practicable steps to prevent prohibited conduct.

Importance

This is particularly valuable to the preventive-liability concept.

It demonstrates a legal model in which responsibility may be connected not only to the harmful act itself but also to the failure to take reasonable preventive steps.

Formula

Known Risk + Ability to Prevent + Failure to Act = Potential Responsibility

This resembles predictive harm-prevention theory more closely than purely compensatory liability.

18. Case Law 9: Khaled Salem Musabeh Humad Al Mheiri v John Cameron — DIFC CA 008/2025

In Khaled Salem Musabeh Humad Al Mheiri v John Cameron [2025] DIFC CA 008, judgment was delivered on 1 September 2026.

The Court of Appeal allowed the appeal in part, set aside the relevant first-instance findings concerning deceit/mistake and remitted the matter for retrial. The judgment also considered an argument concerning future predictions and whether an unfulfilled prediction about future events could itself amount to deceit under UAE law.

Importance

This is particularly relevant to predictive legal theory.

A prediction about the future is not automatically equivalent to a false representation of present fact.

Principle

Prediction itself does not automatically create civil liability.

The legal analysis must distinguish:

opinion;

prediction;

representation of present fact;

fraudulent misrepresentation;

negligent conduct.

This distinction is essential when predictive systems are used in civil transactions.

19. Predictive Harm Prevention vs Ex-Post Liability

20. Does Predictive Prevention Replace Civil Liability?

Strict legal answer: No.

At present, UAE law does not establish a general rule saying:

“If an AI system could have predicted the harm, ordinary civil liability disappears.”

Instead, predictive prevention should be regarded as a complementary layer.

The model becomes:

Stage 1 — Prevention

Identify and mitigate foreseeable risk.

Stage 2 — Intervention

Stop or reduce harmful conduct.

Stage 3 — Investigation

Determine what happened.

Stage 4 — Liability

Identify responsible parties.

Stage 5 — Remedy

Provide compensation, restoration, injunction or another appropriate remedy.

21. Why Complete Replacement Would Be Difficult

A. Prediction is probabilistic

A model may say:

80% risk

but the event may never occur.

B. False positives

The system may identify harmless conduct as dangerous.

C. False negatives

The system may fail to detect a genuine risk.

D. Causation remains important

Prediction does not prove that the defendant caused the eventual damage.

E. Legal duties differ

Not every person who can predict a risk has a legal duty to prevent it.

F. Prevention may itself cause harm

Stopping a transaction or service may cause economic loss.

G. Human judgment remains necessary

Civil liability often depends on context and proportionality.

22. Predictive Harm Prevention and the Precautionary Principle

The concept resembles the broader precautionary principle:

When a serious risk of harm can reasonably be identified, preventive measures may be justified before the harm becomes irreversible.

In civil law, however, precaution must be connected to an actual legal basis.

The court should not impose unlimited responsibility merely because technology makes prediction possible.

23. The “Ability to Prevent” Factor

A particularly important future concept is:

Who had practical control over the risk?

Consider an autonomous delivery system.

The following parties may be involved:

manufacturer;

software developer;

operator;

platform;

maintenance provider;

data provider;

user.

If the AI predicts a dangerous software failure, responsibility should not automatically fall on everyone.

The legal inquiry should examine:

Who received the warning?

Who understood it?

Who controlled the system?

Who could stop it?

Who had a legal duty?

Who ignored the warning?

Did that failure cause the damage?

24. Predictive Harm Prevention and AI

AI makes prevention technologically possible at unprecedented scale.

Examples

Fraud

Suspicious transaction → automatic alert → transaction review

Cybersecurity

Anomalous access → warning → account suspension

Construction

Structural sensor detects abnormal stress → inspection → repair

Medical technology

Predictive system detects dangerous pattern → human review → intervention

Autonomous vehicles

Collision probability detected → emergency braking

Financial technology

Unusual transaction → risk score → human verification

The legal challenge is converting technological capability into a legally reasonable duty without imposing unlimited liability.

25. Predictive Prevention and Standard of Care

The standard of care may evolve as technology becomes more widely available.

Suppose in 2020 a certain risk was difficult to detect.

Suppose by 2030 an inexpensive and reliable predictive system can detect it.

A future legal question may be:

Does the availability of reliable predictive technology affect what constitutes reasonable care?

Potentially, yes—but this would require consideration of:

industry practice;

statutory duties;

contractual obligations;

cost;

reliability;

accessibility;

seriousness of risk;

professional standards.

Technology does not automatically create a duty, but it may influence the assessment of reasonable precautions.

26. Predictive Prevention and the “Duty to Monitor”

A future UAE civil-liability framework could potentially recognise differentiated duties to monitor high-risk systems.

Examples could include:

High-risk AI

Continuous monitoring.

Autonomous machinery

Predictive maintenance.

Financial platforms

Fraud detection.

Medical systems

Safety alerts.

Critical infrastructure

Predictive failure analysis.

The duty should depend on the nature and magnitude of the risk.

27. Predictive Prevention and Proportionality

Prevention should not be unlimited.

Suppose a system predicts a 0.01% risk of minor financial loss.

It may be disproportionate to impose enormous restrictions.

Conversely, suppose the system predicts a significant risk of:

death;

serious injury;

major environmental harm;

systemic financial loss.

The justification for intervention becomes stronger.

Therefore:

Risk magnitude + probability + reversibility + cost of prevention = proportional preventive response

28. Predictive Prevention and Injunctions

Interim injunctions provide an important existing legal analogy.

The Golden Sands decision demonstrates that a court may consider whether damages would be adequate and may preserve a party's position before final trial.

This reflects a broader civil-justice principle:

The legal system does not always have to wait for final harm before acting.

Predictive technologies may provide better factual information for deciding when preventive intervention is appropriate.

29. Predictive Prevention and Restorative Remedies

The current Civil Transactions Law is not restricted to a purely monetary conception of compensation.

Article 256 permits, in appropriate circumstances and upon request, restoration of the previous situation or performance of a specific matter connected with the harmful act as a form of compensation.

This is important conceptually because civil justice can move beyond:

“Pay money after harm.”

toward:

“Restore or correct the harmful situation.”

Predictive prevention takes this one step further:

“Prevent the harmful situation from occurring.”

30. Predictive Prevention and Causation

Causation remains one of the hardest issues.

Suppose:

AI predicted 90% probability of equipment failure.

The operator ignored the warning.

The machine later failed.

The claimant must still establish the legally relevant causal connection.

The following questions arise:

Would the failure have occurred anyway?

Would preventive action have stopped it?

Was the prediction sufficiently reliable?

Was the warning communicated?

Was the operator legally obliged to act?

Did another independent event cause the damage?

Therefore:

Prediction strengthens evidence of foreseeability but does not automatically establish causation.

31. Predictive Prevention and Autonomous Agents

Autonomous AI agents create a particularly difficult problem.

Suppose an AI agent:

negotiates a contract;

makes an investment;

transfers digital assets;

changes software settings;

interacts with another autonomous agent.

If the system causes harm, responsibility may potentially involve:

principal;

developer;

deployer;

operator;

platform;

data provider;

owner.

The legal system may therefore need to focus increasingly on:

Control + Knowledge + Foreseeability + Preventive Capability

rather than simply asking:

Who technically pressed the button?

32. Predictive Prevention and Contractual Allocation of Risk

Contracts may allocate preventive responsibilities.

For example:

“The platform shall monitor abnormal transactions and suspend suspicious transactions.”

If the platform receives a clear warning and fails to act, the contractual obligation may become important in determining responsibility.

However, contractual allocation cannot necessarily override mandatory legal rules or public policy.

33. Predictive Prevention and Evidence

Predictive systems generate valuable evidence:

timestamps;

alerts;

risk scores;

system logs;

audit trails;

warnings;

responses;

intervention records.

This creates a new evidentiary chain:

Prediction → Warning → Knowledge → Opportunity to Act → Failure/Response → Harm

That chain may become important in future civil litigation.

34. Predictive Prevention and AI Accountability

A reliable legal framework should preserve an audit trail.

For example:

Such records can help determine whether a party reasonably responded to a known risk.

35. Predictive Prevention and Algorithmic Errors

The system itself may be wrong.

Therefore, responsibility should also consider:

Model accuracy

How often does the system produce false results?

Training data

Was the dataset reliable?

Drift

Has system performance changed over time?

Validation

Was the model independently tested?

Human oversight

Could an employee override it?

Warning quality

Was the warning understandable?

Response procedure

Was there a reasonable process for acting on warnings?

36. Seven Layers of Preventive Civil Justice

A useful UAE framework is:

Layer 1 — Risk

Identify potential harm.

Layer 2 — Prediction

Estimate probability.

Layer 3 — Warning

Notify the responsible person.

Layer 4 — Prevention

Take reasonable precautions.

Layer 5 — Intervention

Stop or reduce harmful conduct.

Layer 6 — Liability

Determine legal responsibility if harm occurs.

Layer 7 — Remedy

Compensation, restoration, injunction or other appropriate relief.

This shows why prevention does not necessarily eliminate liability.

37. Comparison of the Two Models

Traditional model

Harm → Claim → Proof → Liability → Damages

Predictive model

Risk → Prediction → Warning → Prevention → Monitoring

Integrated UAE model

Risk → Prediction → Prevention → Harm (if unavoidable) → Liability → Remedy

The third model is the most realistic legal development.

38. Important Legal Principle: “Predictive Capacity ≠ Unlimited Duty”

A dangerous consequence of predictive justice would be:

“Because AI can predict something, everyone is legally required to prevent it.”

That would be excessive.

A legal duty should depend on:

applicable legislation;

contractual obligations;

professional standards;

control;

foreseeability;

proximity;

seriousness of harm;

reasonable practicability;

proportionality.

The existence of technology is relevant but not automatically determinative.

39. Case-Law Summary

40. Key Legal Lessons From the Cases

Lesson 1

Courts can intervene before final harm where interim protection is justified.

Lesson 2

Foreseeability is relevant to the existence and scope of duties.

Lesson 3

Technology can improve risk identification.

Lesson 4

AI output requires verification.

Lesson 5

Predictive capability does not automatically establish liability.

Lesson 6

Failure to take reasonable preventive measures can itself become legally significant in appropriate statutory contexts.

Lesson 7

Human responsibility remains central.

41. Future UAE Legal Model

A future UAE framework for high-risk AI and autonomous systems could potentially use:

A. Mandatory risk assessment

Before deployment.

B. Continuous monitoring

During operation.

C. Automated alerts

When risk thresholds are exceeded.

D. Human override

Where serious consequences are possible.

E. Incident reporting

After a harmful event or near miss.

F. Audit logs

To establish who knew what and when.

G. Preventive injunctions

Where serious future harm is threatened.

H. Ex-post liability

Where prevention fails.

42. Exam-Oriented Answer

If asked:

“Explain whether predictive harm prevention can replace ex-post civil liability under UAE law.”

A strong answer should state:

Predictive harm prevention represents a movement from retrospective compensation toward prospective risk management. UAE law, however, does not presently abolish or replace ex-post civil liability with predictive systems. The current Civil Transactions Law continues to recognise compensation for harm, while civil procedure and interim-relief mechanisms can provide protection before final adjudication. The emerging technology jurisprudence of the DIFC demonstrates that AI may assist in identifying risks and managing evidence, but human legal responsibility, foreseeability, causation and judicial assessment remain essential.

43. One-Line Exam Answer

Predictive harm prevention in UAE civil law seeks to identify and mitigate foreseeable harm before it occurs, but it presently operates as a complement to, rather than a complete replacement for, ex-post civil liability and compensation.

44. Memory Formula

“P-W-A-L-R”

P — Predict the risk
W — Warn the responsible person
A — Act to prevent harm
L — Liability if harm nevertheless occurs
R — Remedy through compensation/restoration/injunction

Final formula:

Predict → Warn → Prevent → If Harm → Liability → Remedy

45. Conclusion

The movement from ex-post liability toward predictive harm prevention represents a major theoretical change in civil law.

Traditional civil liability asks:

“What happened, who caused it, and how should the victim be compensated?”

Predictive civil justice asks an additional question:

“Could the risk have been identified and reasonably prevented before the harm occurred?”

UAE law already contains legal concepts that make this transition possible: foreseeability, duties of care, interim injunctive protection, preventive obligations in particular statutory contexts, restoration-oriented remedies and sophisticated digital-evidence mechanisms.

The cases of Golden Sands, Faizal Babu Moorkath, Haya Spa, Gate Mena, AES, Klesta Eshja, Stelian Gheorghe and Niran collectively illustrate these different components. They do not, however, establish a general UAE rule that predictive AI replaces conventional civil liability.

The legally safer and more accurate conceptual model is:

Prediction does not eliminate liability; it changes when and how the legal system can intervene.

Thus, the future UAE model may progressively move from:

“Compensate after harm”

toward:

“Predict risk → prevent harm → intervene early → compensate only where prevention fails.”

That would represent an evolution of civil liability from a primarily reactive model toward a preventive and risk-governance model, while retaining judicial control, proportionality, causation and individual legal responsibility.